Who Supports Suicide Terrorism in Bangladesh? What the Data Say

Politics and Religion, page 1 of 40, 2017. © Religion and Politics Section of the American Political Science Association, 2017 doi:10.1017/S1755048317000347 1755-0483/17

C. Christine Fair, Ali Hamza, and Rebecca Heller Georgetown University

Abstract: Bangladesh, one of the world’s largest Muslim countries, is generally viewed as a success story with a strong tradition of secular democracy. Unfortunately, this assertion rests on a weak empirical foundation. Since becoming independent from Pakistan in 1971, democracy and secularism have been consistently undermined. Moreover, since 2000 Bangladesh has experienced more than 100 incidents of Islamist terrorism. More recently, the Islamic State and Al-Qaeda Indian Subcontinent have assaulted religious and ethnic minorities as well as secular and liberal activists. While these turns of events are alarming, Islamist militancy in Bangladesh remains understudied. In this article, we address this lacuna by undertaking regression analysis of recent Pew Research Center survey data to exposit the determinants of popular support of Islamist terrorism. With this study, we hope other scholars will be motivated to turn their attention to this increasingly important state.

1. INTRODUCTION: ISLAMIST MILITANCY IN BANGLADESH

In March 2014, Dan Mozena, then the United States Ambassador to Bangladesh, said that Bangladesh is “a moderate and generally secular and tolerant — though sometimes this is getting stretched at the moment — alternative to violent extremism in a very troubled part of

We are very grateful to Rebecca Littman, Elizabeth Nugent, Jacob N. Shapiro, and Neil Malhotra whose collaborations on Pakistan-related projects yielded important insights for this paper. We are also thankful to Ali Riaz and Alyssa Ayres for their comments on an early draft of this essay as well as the trenchant criticisms offered by the various anonymous reviewers. Despite the efforts and interventions of these varied persons, we alone are responsible for any errors of fact or interpretation. Finally, this project could not be possible with the support of the Georgetown Undergraduate Research Opportunities Program, which sponsored the work of Ms. Heller, or the unstinting support of the Security Studies Program, which supported the work of Mr. Hamza. Address correspondence and reprint requests to: C. Christine Fair, Georgetown University, Security Studies Program in the Walsh School of Foreign Service, 202 Mortara Building, 37th and O Streets, NW, Washington, DC 20057. E-mail: [email protected].

the world” (Gowen 2014). While Mozena’s statement reflects the general perception that Bangladesh is a success story of a moderate, secular, Muslim democracy, this view never rested on strong empirical ground. Indeed, since Bangladesh’s independence from Pakistan in 1971, the dura- bility of both secularism and democracy has been undermined by numerous military coups — many of which involved multiple counter-coups before a clear “victor” emerged — in 1974–1975, 1977–1980, 1981– 1982, 1996, and 2007. (In 2012, the military announced that it had thwarted yet another coup in January of that year.) Bangladesh’stwo mainstream political parties, the left-of-center Awami League led by Prime Minister Sheikh Hasina and the right-of-center Bangladesh National Party (BNP) led by former Prime Minister Khaleda Zia, are known more for their rivalry, corruption, and incompetence than for good governance. Since independence, Bangladesh has experienced creeping Islamism that enjoys popular support (Fair and Oldmixon 2015). Increasingly, Bangladesh is the site of Islamist violence. According to data derived from the Global Terrorism Database maintained by the National Consortium for the Study of Terrorism and Responses to Terrorism (START), there have been 944 terrorist attacks between 2000 and 2015. While the perpetrators of the majority of events in the Global Terrorism Database are listed as “unknown” or attributed to political parties, 114 were perpetrated by confirmed Islamist militant groups. However, the victim yields for these attacks are quite low with an average of one fatality and under seven persons wounded per attack. These low victim yields may explain in part why this kind of violence in Bangladesh has attracted little attention. By way of contrast, using the same dataset, the Lashkar-e-Taiba, in the same period, has perpetrated 136 attacks killing about six persons and wounding another 14 per attack in India. Perhaps if Bangladeshi terrorist groups were more lethal, they would garner more attention. More recently, the Islamic State (IS) and Al-Qaeda Indian Subcontinent (AQIS) have claimed many of these attacks (Kumar and Iyengar 2016), casting a further pall over Bangladesh’ ostensible success story. We aver that Bangladesh merits greater scholarly attention. After all, Bangladesh has one of the world’s largest Muslim populations with more than 150 million Muslims.1 Bangladesh’s Muslim population exceeds the combined populaces of Iran (81.2 million), Afghanistan (32.6 million), and Saudi Arabia (27.7 million) (U.S. Central Intelligence Agency 2016a; 2016b; 2016c). But it is also one of the world’s least developed countries: Bangladesh ranks 142 out of 190

Downloaded from https:/www.cambridge.org/core. IP address: 73.180.230.14, on 28 Jun 2017 at 15:38:25, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/S1755048317000347 Who Supports Suicide Terrorism in Bangladesh? 3

countries according to the United Nations Human Development Index (United Nations Development Program 2015). Its citizens also view their country as plagued by corruption, ranking 139 out of 167 countries ranked in Transparency International’s 2015 Corruption Perception Index (Transparency International 2016). Bangladesh is an important provider of global security, consistently being one of the largest contributors to United Nations Peacekeeping Missions (United Nations 2016).2 While not a top- tier military, its military forces are ranked 52 out of 126, using an index that considers the forces’ end-strength, diversity and number of weapons systems as well as measures of national power (Global Fire Power 2016).3 Whereas the origins of Islamist militancy and support for the same are fairly well characterized in other Muslim countries, these issues have not been explored empirically in Bangladesh. To address this lacuna, we employ data for Bangladesh from the Pew Research Center’s World’s Muslims Data Set to conduct an ordinary least squares regression analysis on Bangladeshi support for suicide terrorism using a set of theoretically derived independent and control variables. We find that support for suicide attacks is rather high in Bangladesh with almost half of the population finding them justified in some measure. Levels of justification for suicide attacks in Bangladesh are considerably higher than in most other Muslim countries studied by the Pew Research Center. We find that respondents who support literal interpretations of the Quran, including physical criminal punishments (known as Hudood punishments), are more likely to support suicide terrorism whereas respondents who support traditional roles of Muslim leaders (e.g., settling family disputes) are less likely to do so. We also find that males and those who view themselves as economically well-off are more supportive of suicide attacks, whereas those who are better educated are less supportive. We organize the remainder of this article as follows. In Section 2, we provide an overview of the Islamist militant landscape in Bangladesh. In Section 3, we review the theoretical and empirical literature on the determinants of support for Islamist violence from which we draw out several testable hypotheses. In Sections 4 and 5, respectively, we describe the data and methods that we employ in this analysis and present our empirical findings. In Section 6, we conclude with a discussion of the implications of this research.

2. ISLAMIST MILITANT MILIEU IN BANGLADESH

Three complex international developments have enabled the growth of Islamist militancy in Bangladesh. First, during the 1980s, some Bangladeshis participated in the so-called “jihad” to oust the Soviets from Afghanistan where they learned to fight. Returning militants brought with them their new knowledge of insurgent warfare and jihadist ideology to Bangladesh (Hasan 2011). Bangladesh received about 15% of its bilateral development assistance aid from the United States in the 1980s and because the United States vigorously supported the efforts in Afghanistan (along with Saudi Arabia and China), Bangladesh tolerated its citizens traveling to and from Afghanistan to fight (Hasan 2012). Second, in the early 1980s, Muslim ethnic Rohingyas (International Crisis Group 2014) formed the Rohingya Solidarity Organization (RSO) in the wake of a massive military operation waged by the Myanmar military that drove some 200,000 Rohingya into Bangladesh.4 The RSO enjoyed explicit support from the Bangladesh Jamaat-e-Islami Bangladesh (BJeI), which is Bangladesh’s largest Islamist party which aims to transform Bangladesh into an Islamic state. While the size of the RSO remains debated, analysts assess that “small numbers” of Rohingya militants continued to train in remote bases in Bangladesh oppo- site Myanmar’s Maungdaw district until the 1990s (Marshall 2014; International Crisis Group 2014). Rohingyas became sources of recruitment for different Islamist militant groups including the Afghan Taliban and al-Qaeda, detailed below (Lintner 2002a; 2002b; 2003). Three, as we describe below, Bangladesh became one of the regional hubs which Pakistan has used to train, hide, and dispatch Islamist terror groups into India for well over a decade (Montero 2010). There are several Islamist militant groups that have operated in the recent past or are currently operating in Bangladesh, some of which are strictly domestic while several operate across South Asia and beyond. With varying degrees of evidence, some scholars and the current Bangladeshi government assert that the BJeI is the taproot of many of the country’s Islamist militant organizations. The BJeI has been able to develop political influence disproportionate to its share of votes because of its association with the BNP and its role as kingmaker in helping the BNP create a majority coalition in 2001 (Fair and Oldmixon 2015). The BJeI has attracted episodic international scrutiny since 2001 due to its purported deep involvement in numerous terror attacks which targeted Hindus, Ahmedis, Awami League activists and liberal activists in

Bangladesh. The student wing of the BJeI, Islami Chhatra Shibir or Jamaat-Shibir, has also been involved in militant attacks (Kumar 2014).5 The Jamatul Mujahideen Bangladesh (JMB) is an indigenous, Bangladesh-based militant group which coalesced in 1998. The JMB perpetrated many well-publicized terrorist attacks in the early 2000s, including a shocking August 2005 attack in which the group set off 459 bombs simultaneously in 63 of Bangladesh’s 64 districts in effort to push the country into adopting Sharia law (The Indian Express 2015). JMB has also been linked to recent violence in Bangladesh, including an incident in Dinajpur at the end of 2015 in which an Italian Catholic priest was attacked (BBC News 2016a). Harkat-ul-Jihad-al Islami Bangladesh (HuJI-B) was founded in 1992 and facilitated the development of many other Islamist groups in the country (Riaz and Fair 2011). HuJI-B is widely believed to be behind some of the earliest Islamist terrorist actions in Bangladesh, including the 1993 death threats against the feminist author Taslima Nasreen, who had to leave Bangladesh after a $5,000 bounty was put on her head (Lintner 2003). Analysts believe that HuJI-B tried to assassinate both Shamshur Rahman, a famed secular poet, and Prime Minister Sheikh Hasina (Hasan 2011). Many of HuJI-B’s members came from or were trained by foreign militants, especially fighters from the war in Afghanistan (U.S. Department of State Office of the Coordinator for Counterterrorism 2006). In its early years, Osama bin Laden funded the group (Hasan 2011). Many Rohingyas also closely collaborated with and even trained HuJI- B members in the 1990s (Riaz and Fair 2011). HuJI-B recruited members from Rohingya communities in southeastern Bangladesh (Lintner 2002a; 2002b). Because of their abject poverty and the limited number of options available to them, Rohingya fighters were often assigned to the riskiest fighting jobs, doing tasks such as carrying equipment or removing mines (Lintner 2002a; 2002b). The relationship between the RSO and Bangladeshi Islamist groups like HuJI-B and the JMB ultimately proved beneficial to both sides: the JMB, for example, taught Rohingyas to build and detonate bombs, while Rohingya experts trained JMB members in arms usage (Riaz and Fair 2011). Riaz and Fair argue that “this relationship, tacitly encouraged by the then Bangladesh government headed by Khaleda Zia, helped shape future relationships between various militant groups who use the hill tracts as their bases” (Riaz and Fair 2011, 158). Some of these fighters stayed in Bangladesh to work for HuJI-B or JMB, but many went to war to aid al-Qaeda and the Taliban in

Afghanistan (Lintner 2002a; 2002b). The presence of the camps has allowed the different militant groups to collaborate, train together, and strengthen each other. Another group of concern is Hefazat-e-Islami (HeI), which is “an alliance of teachers and students associated with radical madrasas and with JeI” (Stark 2015). It has ties to more than 25,000 madrasahs in Bangladesh (Stark 2015; Mustafa 2013) and has linkages to the Afghan Taliban (Bashir 2015). This amalgamation of several Islamist organizations came to the fore when its members participated in large public dem- onstrations supporting the murder of a young, secular blogger in 2013. It drew further attention when it released its 13-point charter of demands, which included enactment of an anti-blasphemy law with provision for the death penalty, imposing restrictions on women, declaring Ahmedis as “non-Muslims,” among others (Mustafa 2013). In addition to these Bangladeshi groups, there are several regional Islamist militant groups that operate in Bangladesh. One such group is Lashkar-e-Taiba (LeT), which is based in Pakistan but also operates in Bangladesh. LeT has organized many terrorist attacks in both Bangladesh and India, though it is most renowned for its November 2008 assault on multiple targets in the Indian mega-city of Mumbai. LeT also operates under several front organizations, including Jamaat- ud-Dawa and Filah-e-Insaniat Foundation, all of which have been pro- scribed by the United Nations as “terrorist organizations” (U.S. Department of State 2010). Mufti Obaidullah, a Bangladeshi national who was associated with LeT and had fought in Afghanistan four times — in 1988, 1990, 1991, and 1992 — was arrested in 2009 after spending 14 years teaching in a Bangladeshi madrassah under a fake name (The Daily Star 2009; Habib 2009; The Hindustan Times 2009). Two transnational groups, (AQIS and the IS), have also become increasingly active in Bangladesh in recent years. AQIS has taken responsibility for a number of murders, including murders of secular publishers and bloggers, at least one of whom was American (Barry and Manik 2015), while IS has taken responsibility for other attacks on foreigners, homosexuals, Shia, Ahmadis, Sufis, and religious minorities among others (Bashir 2015; Manik and Barry 2016; Dwibhashyam 2016;Al Arabiya English 2015). In July 2016, several Bangladeshi terrorists attacked Dhaka’s Holey Artisan Bakery killing 20 persons over the course of more than 12 hours (BBC News 2016b). The attackers were in touch with the IS and dedicated the attack in their name, although

there is no evidence that the IS aided or abetted the terrorists in question (Swami 2016). Islamist militants have targeted secular writers and bloggers in particular, with a “hit list” available online. The Ansarullah Bangla Team, a banned Bangladeshi Islamist group that first gained attention in 2013, is responsible for many of these murders. Ansarullah Bangla Team, which has also called itself Ansar al-Islam and Ansar Bangla 7, is affiliated with Al-Qaeda (Al-Mahmood 2015). Dozens of Bangladeshis, including persons of Bangladeshi extraction in the United Kingdom, have gone to fight with the IS (Dodwell, Milton, and Rassler 2016; Roul 2016a; 2016b; Hussain 2016). Recently, IS’ English-language magazine Dabiq offered a tribute to a Bangladeshi militant who died in Syria (Islamic State1437 (Rajab); Khandake 2016). It appears as if AQIS and IS are com- peting for influence, people, and cash in Bangladesh (Mahmood 2016; Roul 2016b). In attempt to retain its stronghold over South Asia, AQIS has launched its own efforts to focus upon Bangladesh and other parts of South Asia (Fair 2014; Chandran 2015). Despite these various claims by IS and AQIS, Bangladesh’s government has generally insisted that these groups do not have a presence in the country (Ganguly 2016; Agrawal 2016) with the notable exception of the Holey Bakery attack, which forced the government to niggardly concede the potential for international connections. Instead, Prime Minister Sheikh Hasina alleges that the BNP and BJeI are conducting these attacks “to destabilize the country” (Manik 2016). Despite Hasina’s ostensibly secular reputation, she has demurred from explicitly condemning the killing of secular activists and minorities and has even blamed the victims for provoking the terrorists with their controversial speech (Sing 2016).

3. REVIEWING THE EXTANT LITERATURE: EXPLAINING SUPPORT FOR ISLAMIST MILITANCY

The body of scholarly literature examining support for violent groups has traditionally focused on ethnic conflicts (Horowitz 1985), state repression (Scott 1976), grievances (Gurr 1970), and a variety of individual-level explanatory factors and motivations (inter alia Blair et al. 2013; Blair, Lyall, and Imai 2013; Chiozza 2007; Tessler and Robbins 2007; Shafiq and Sinno 2010; Fair et al. 2016). Here we review the expansive literature on individual-level explanations for support for militant violence and

consider these studies with reference to Bangladesh’s political history with the intent of generating several hypotheses which we subsequently test empirically.

3.1. Poverty and Support for Violence

One sort of personal motivation derives from poverty or perceived poverty. Several scholars have argued that low-income individuals are more likely to support militant organizations due to feelings of powerless- ness and general dissatisfaction with the current political system (see, Abadie 2006; Esposito and Voll 1996; Piazza 2007; Tessler and Robbins 2007). Yet another possible mechanism for the relationship between poverty and support for violence focuses on opportunity costs: persons living in poverty have lower opportunity costs associated with supporting political violence than their wealthier counterparts, making them more likely to do so ceteris parabis (Gates 2002; Dal Bó and Dal Bó 2011). Ethan Bueno de Mesquita offers a somewhat different interpretation of the opportunity cost argument. He observes that poorer countries are more likely to produce terrorists but notes that the terrorists themselves are generally better educated than the societies from which they are recruited. This is because terrorist groups, like any other hirer of labor, recruit on quality: as long as there are more high quality recruits than are lower quality persons willing to join, militant groups will hire the higher quality individual. There are two components to his explanation. First, during economic downturn, those better educated persons who find themselves unemployed face larger negative changes in their opportunity costs to participate in militancy than do persons with lower educational attainment rendered unemployed. (He uses education as a proxy for quality or human capital.) Second, terrorist groups select on quality when they can and economic downturn increases the proportion of higher quality recruits compared to periods of economic growth (Bueno de Mesquita 2005). The empirical evidence on the relationship between poverty and support for political violence is mixed (Burgoon 2006; Krueger and Malekova 2003; Berrebi 2007). Tessler and Robbins find that “neither personal nor societal economic circumstances, by themselves, are important determinants of attitudes toward terrorism directed at the United States” (Tessler and Robbins 2007). Shafiq and Sinno, using Pew’s Global Attitudes Survey from 2005, find that the relationship between education

and income on the one hand and support for suicide bombings on the other varies across countries and targets (Shafiq and Sinno 2010). Chiozza, also employing Pew’s Global Attitudes Survey data, similarly concluded that individual-level income and support for suicide bombing varies across countries (Chiozza 2011). Mousseau (2011), mobilizing Pew’s 2002 Global Attitudes Survey data from 14 Muslim nations, observed that support for Islamist terrorism is highest among the urban poor. In 2009, Blair and co-workers fielded a 6,000-person representative survey in the four main provinces of Pakistan. In contrast to the works cited above, which measures support for militancy using direct question- ing, they used a series of endorsement experiments (Bullock, Imai, and Shapiro 2011) to discern indirect support for a variety of Islamist militants in Pakistan (Blair et al. 2014). They found that poor Pakistanis were most opposed to the militants than were middle-class citizens and that this aver- sion was strongest among the urban poor, particularly those in violence- afflicted districts. They presented evidence that suggested urban poor are most opposed to these groups because they are most exposed to the negative externalities of militant violence. Following up on that work, Fair and co-workers fielded a nationally-representative survey in Pakistan among 16,279-persons. Per Blair et al. (2013), they employed the endorsement experiment methodology to measure indirect support for several Islamist militant groups operating in or from Pakistan. They found, using expenditures as a measure of socio-economic standing rather than income, that lower-class respondent were less supportive of militant groups. To further explore the linkages between socio-economic status and support for Islamist militancy, they experimentally induced per- ceptions of relative poverty among half of their respondents. They found that support for Islamist militant groups was lowered among those in the treatment group, relative to the untreated group (Fair et al. 2016). At first blush, this work in Pakistan seems most germane to our present study of Bangladesh rather than similar work on Middle Eastern or Southeast Asian polities for several reasons. First, Pakistan and Bangladesh were the same country until 1971. Prior to that the peoples of South Asia shared hundreds of years under British governance, first under the British East India Company (1600–1857) and then under the Raj (1857–1947) as well as centuries under various Portuguese, French, Mughal, Hindu, and Buddhist governance structures prior to the arrival of the British. The citizens of today’s Bangladesh and those of today’s Pakistan, as well as those in many states in north India, embrace a

common socio-cultural history (van Schendel 2009). Second, Bangladesh exhibits many — but by no means all — of the sectarian divides (i.e., Sufi, Ahl-e-Hadis, Deobandi, Shia, Ahmedi, etc.) and religious traditions of Pakistan stemming from the afore-noted shared social, cultural and religious history. Third, Islamist organizations such as JeI developed off- shoots in Bangladesh from their parental organizations in Pakistan, which in turn are rooted in 19th century Islamist political movements in pre-independence India (Titus 1959; Eaton 1993; Robinson 2004). Finally, there has been considerable movement of Islamist militant groups between both Pakistan and Bangladesh as noted above (Riaz 2007; Jalal 2008). However, there also critical ways in which Bangladesh does not resem- ble Pakistan. First and foremost, the Bangladeshi state has never actively promoted jihad nor has the state nurtured and deployed jihadi organizations as a tool of statecraft. Second, unlike Pakistan where militant groups openly operate and roam freely, Bangladesh is a hostile environ- ment and groups are forced to operate through clandestine networks and are the subject of ongoing law enforcement activities. Third, Bangladesh’s popular culture does not valorize Islamist “freedom fighters” as is the case in Pakistan. In fact, in Bangladesh “freedom fighters” refer to those Bengalis who opposed Pakistani state and its efforts to Islamize ethnic Bengalis (Riaz 2016). Finally, Bangladesh differs from Pakistan in that while Islamist violence has increased in recent decades, it is nowhere as lethal as it is in Pakistan. Thus, findings of Blair et al. (2013) that poorer persons in Pakistan experience the negative externalities of terrorist violence and thus less receptive to their appeals is not likely to apply here. Given the inconsistent empirical findings on the relationships between economic standing and support for violent politics, there is no a priori prediction for what relationship may obtain in Bangladesh. We derive two testable null hypotheses from this literature:

H1: Economic status is not correlated with support for Islamist violence.

H2: Perceived economic status is not correlated with support for Islamist violence.

3.2. Clash of Civilizations

Another explanation that scholars have mobilized to explain support for Islamist violence is the “clash of civilizations” thesis outlined by Huntington (1984; 1993; 1996), which argues for a fundamental conflict between the Christian West and the so-called Islamic World, to posit that support for terrorism and/or militancy derives from adherence to Islam itself (Laqueur 1999; Calver 2002; Stern 2003; Mendelsohn 2005). Accordingly, several public intellectuals argued that support for Islamist militancy is rooted in Muslim religiosity or faith (Laqueur 1999; Calver 2002; Stern 2003; Mendelsohn 2005). While there is limited anecdotal as well as empirical support for the asserted causal relationship between Islamic piety and political violence (Weinberg, Pedazhur, and Canetti- Nisim 2003; Ginges, Hansen, and Norenzayan 2009), most analytic studies find little association between simply believing in Islam and supporting violent politics (Tessler and Robbins 2007; Esposito 2002; Tessler and Nachtwey 1998). When a correlation between embracing Islam and violence does exist, the relationship is driven by a particular understanding of Islam (e.g., for example, beliefs about the efficacy or compulsory nature of individual militarized jihad) (Fair, Malhotra, and Shapiro 2014; Fair, Littman, and Nugent 2017). Other studies have found that adherence to specific sectarian traditions predict support for Islamist militant groups (Fair 2015).6 At least two scholars have presented limited evidence that individuals with greater knowledge of Islam, obtained through Quranic study groups and other pietic practices, are better able to resist the arguments of militant thought leaders and thus less likely to support Islamist militant politics (Wiktorowicz 2005; Fair, Goldstein, and Hamza 2016). This set of studies gives rise to a third testable hypothesis, namely:

H3: Piety is not correlated with support for Islamist violence.

Another argument that epistemologically draws from the so-called “Clash of Civilizations” thesis is the contention that preferences for political Islam or Islamism — terms that scholars tend to use interchange- ably — explains support for Islamist violence. Studies exploring these purported links have yielded contradictory conclusions. Fair, Littman, and Nugent contend that these conflicted results likely stem from the fact that there is no universally held understanding of what the application of Sharia looks like and from the problematic survey items that analysts

use to proxy support for “political Islam” (Fair, Littman, and Nugent 2017). They identify and tweeze out several factors that are usually col- lapsed into the concept of “support for Sharia.” Motivated by the notion that some Muslims view Islamic government in terms of good governance in Afghanistan, Pakistan, the Caucasus and Central Asia, Egypt, Iran, and Turkey (Abdul-Ahad 2008; Collins 2007; Rheault and Mogahed 2008; Fair, Littman, and Nugent 2017), they developed several survey items to specifically indicate support for Sharia as a vehicle for providing government services and another set of questions that elicit support for scriptural literalism (physical punishments, restrictions upon women). They construct index variables for both notions of “provision” and “punishment” and find that the former does explain support for Islamist militancy while the latter does not. The mechanism they posit is that Islamist militants specifically argue for the kind of physical punishments included in the index while tending not to argue for the kinds of actions included in the other index.7 In the context of Bangladesh, Bangladeshi militants also make specific appeals to Hudood punishments as they do elsewhere and they have specifically sought to use violence to coerce the state to abandon its judicial system and replace it with sharia courts (Bhattacharyya 2015). In contrast, Bangladesh’ Islamist militant groups have not made any arguments that Sharia involves the provision of public services. Applying this component approach to Sharia to Bangladesh gives rise to two additional testable hypotheses, namely:

H4: Persons who are favorable to scriptural literalism and physical punishments will be more likely to support Islamist political violence.

H5: Support for the notion of provision of services will not be related to support for Islamist violence.

While militant groups in Bangladesh have been arguing for a government based upon their interpretation of Sharia, Bangladesh’s left-of- center political elites have demurred from such debates about the role of secularism in the state and the current government has astutely avoided any direct national conversation even though Bangladesh’s civil society has strong and vocal proponents of secularism — many of whom have paid with their lives. During Bangladesh’s independence movement, the West Pakistan army and their JeI allies specifically singled out and

murdered Bangladesh’s secularists who today are revered as Bangladesh’s heroes (Bass 2013). As noted above, Islamist militants have specifically targeted Bangladesh’s secularists and liberal activists presumably as they oppose the Islamist agenda of rendering Bangladesh an Islamist state. The historical role that the struggle over secularism has played in Bangladesh’s civil and not-so-civil society (Khondker 2006), gives rise to our final hypothesis, namely:

H6: Respondent support for secularism should be negatively related to respondent support for Islamist militancy.

4. DATA AND METHODS

We leverage publicly available data for Bangladesh from the Pew Foundation’s “World’s Muslims Data Set.” These data are derived from a nationally representative survey of 1,918 adult respondents, conducted between November 2011 and February 2012 in Bangladesh’s national language, Bangla (Bengali). Pew’s sampling design was based on “stratified area probability of all seven administrative divisions to population size and urban/rural population” (Pew Research Center 2013a; 2013b). Pew included Bangladesh in its Global Attitudes Survey in 2014; however, that sample size was smaller (1,000 persons) and that survey did not include the variables that we require for this analysis. While this dataset is the largest such dataset on Bangladesh and includes the most comprehensive assessment of Bangladeshi religious beliefs and practices, it has a number of drawbacks. First, Pew does not provide any geographic information other than a binary of whether or not the respondent lives in a rural or urban area. It would have been ideal had Pew at least provided information about the division in which respondents live. (Bangladesh has eight divisions.) With this kind of information, we could have merged in extant data on terrorist events (Global Terrorism Data Base) to exploit geographical variation among our respondents as a potential explanatory factor in support for violence per Blair et al. (2013). Second, the questions that Pew employs in its questionnaire are not tailored to Bangladesh; rather, Pew generally poses identical questions in all countries the firm surveys. Third, Pew’s instrument does little to illuminate how Muslims in their multi-country study understand important yet complex concepts like “Sharia.” Finally, Pew uses a standard

question about support for suicide attacks that are denuded of context even though previous studies show that respondent support for suicide attacks is highly sensitive to context and details of the attackers and victims.8 Despite these — and other — shortcomings, these are the best data for this study. We derive our dependent variable from the sub-optimal question that asked respondents: “Some people think that suicide bombing and other forms of violence against civilian targets are justified in order to defend Islam from its enemies. Other people believe that, no matter what the reason, this kind of violence is never justified. Do you personally feel that this kind of violence is often justified to defend Islam, sometimes justified, rarely justified, or never justified?” Despite the problematic nature of this question, the vast majority of respondents answered it. Fewer than 3% of the sample indicated that they “don’t know” or “refused.” We treat these observations as missing in our regression models. To test the posited relationship between actual economic status and support for Islamist violence (H1), we use the question that asks respondents about monthly income and use the categories provided by Pew (Tables 1a, 1b). This variable is not ideal for two reasons. First, persons are frequently not honest when asked to report income. For this reason, it is preferable to ask respondents about expenditures. Second, Pew has provided this variable in pre-set bins whereas we would prefer an continuous numerical income variable with which we could construct our own categories (e.g., quintiles). To test the posited relationship between perceived economic status and support for Islamist violence (H2), we use a question that asks respondents: “And what about your personal economic situation, how would you describe it — is it very good, somewhat good, somewhat bad or very bad?” To test our hypothesized relationships in H3 and H4, we constructed two indices following the methods adopted by Fair, Littman, and Nugent (2017). To test H3, we constructed a simple additive index, called “piety,” from five questions about religious beliefs and specific practices with a possible range of zero (least pious) to one (described in Appendix 1). Because this and our other additive two indices are positive measures, we treated “Don’t Know” and “Refused” responses as zeros in tabulating the indices’ values. Factor analysis (shown in Table 2) confirms that this index taps into two specific concepts of belief and practice. To test H4 we create a variable called “Hudood,” which resembles the “imposes” of Fair, Littman, and Nugent (2017).9 This variable, scaled

https:/www.cambridge.org/core Table 1a. Descriptive statistics

https:/www.cambridge.org/core/terms Frequency Percentage Percentage (without (without (with Categories weights) weights) weights*)

Dependent Variable

Q89: Some people think that suicide bombing and other forms of Often Justified 159 8.29% 9.22%

. IPaddress: violence against civilian targets are justified in order to defend Islam from its enemies. Other people believe that, no matter what the reason, this kind of violence is never justified. Do you personally feel that this kind of violence is often justified to 73.180.230.14 . defend Islam, sometimes justified, rarely justified, or never https://doi.org/10.1017/S1755048317000347 justified? Sometimes Justified 331 17.26% 17.09% Rarely Justified 338 20.23% 20.04% , on Never Justified 985 51.36% 50.96%

28 Jun2017 at15:38:25 Don’t know 42 2.19% 1.96% Refused 13 0.68% 0.73% Total 1,918 100% 100% Independent Variables and Control Variables Male (1 if male) Female* 933 48.64% 47.41% Male 985 51.36% 52.59% Total 1,918 100% 100% , subjectto theCambridgeCore Perceived economic status (Q7: what about your personal Very Good 258 13.45% 14.05% economic situation, how would you describe it) Somewhat good 1072 55.89% 55.27% Somewhat bad 426 22.21% 22.36% Very bad 155 8.08% 7.93% Continued terms ofuse,available at Downloaded from 6Fi,Hma n Hell and Hamza, Fair, 16 Table 1a. Continued https:/www.cambridge.org/core Frequency Percentage Percentage https:/www.cambridge.org/core/terms (without (without (with Categories weights) weights) weights*)

Dependent Variable

Don’t know 5 0.25% 0.32% Refused 2 0.10% 0.07% . IPaddress: Total 1,918 100% 100% Actual Economic Status (q102ban: actual income) Tk. 999 or below 2 0.10% 0.08% Tk. 1,000–1,999 8 0.42% 0.49% –

73.180.230.14 Tk. 2,000 2,999 31 1.62% 1.93% . https://doi.org/10.1017/S1755048317000347 Tk. 3,000–3,999 65 3.39% 2.80% Tk. 4,000–4,999 115 6.00% 5.88% Tk. 5,000–5,999 166 8.65% 8.63% – , on Tk. 6,000 6,999 178 9.28% 8.62% Tk. 7,000–7,999 182 9.49% 9.45% 28 Jun2017 at15:38:25 Tk. 8,000–8,999 206 10.74% 10.50% Tk. 9,000–9,999 348 18.14% 18.48% Tk. 10,000–19,999 451 23.51% 23.88% Tk. 20,000 or above 141 7.35% 8.23% Don’t know 4 0.21% 0.28% Refused 21 1.09% 0.74%

, subjectto theCambridgeCore Total 1,918 100% 100% Level of Education No formal Education 604 31.49% 31.59% Some Primary education 173 9.02% 9.22% Primary Education 151 7.87% 7.79% (completed class V) Some Secondary 332 17.31% 17.62% education terms ofuse,available at Downloaded from h uprsSiieTroimi agaeh 17 Bangladesh? in Terrorism Suicide Supports Who https:/www.cambridge.org/core Table 1a. Continued https:/www.cambridge.org/core/terms

Frequency Percentage Percentage (without (without (with Categories weights) weights) weights*)

Dependent Variable

. IPaddress: Secondary education 244 12.72% 12.56% (completed SSC) Some Higher Secondary 93 4.85% 4.56%

73.180.230.14 Education . https://doi.org/10.1017/S1755048317000347 Higher Secondary 189 9.85% 9.94% Education Some University 60 3.13% 3.24%

, on Education University degree 65 3.39% 3.13% 28 Jun2017 at15:38:25 Post-graduate degree 6 0.31% 0.32% Vocational/diploma 1 0.05% 0.04% Total 1,918 100% 100% Family law (q92a: giving Muslim leaders and religious judges the Favor 1,370 71.43% 71.33% power to decide family and property disputes) Oppose* 474 24.71% 24.92%

, subjectto theCambridgeCore Don’t know 56 2.92% 3.03% Refused 18 0.94% 0.72% Total 1,918 100% 100%

*Indicates the reference category in the regressions. terms ofuse,available at Downloaded from 8Fi,Hma n Hell and Hamza, Fair, 18 https:/www.cambridge.org/core https:/www.cambridge.org/core/terms . IPaddress: Table 1b. Descriptive statistics (with weights) 73.180.230.14

. Mean Stand dev. Range Percentiles https://doi.org/10.1017/S1755048317000347 10% 25% 50% 75% 90% Age 35.49 12.91 18–80 20 25 35 45 55

, on Hudood Index 0.580 .2521 0–1 0.167 0.333 0.5 0.833 0.833 –

28 Jun2017 at15:38:25 Secularism Index 0.3243 0.2090 0 1 0.125 0..25 0.25 0.375 0.625 Religiosity Index 0.781 0.1613 0.07–1 0.55 0.68 0.82 0.9 0.95 , subjectto theCambridgeCore terms ofuse,available at Downloaded from h uprsSiieTroimi agaeh 19 Bangladesh? in Terrorism Suicide Supports Who https:/www.cambridge.org/core https:/www.cambridge.org/core/terms Table 2. Principle component analysis: Rotated factor loadings* (pattern matrix)

Index Factor Factor Factor Factor Factor Name Variable 1 2 3 4 5

Hudood q79a (favor/oppose making the Sharia the official law of the land) 0.5343

. IPaddress: q66 (Sharia is the revealed word of God OR Sharia is developed by men, 0.6811 based on the word of God) q67 (Sharia should be open to multiple interpretations or there is only one, 0.6124 true understanding of the Sharia) 73.180.230.14 . https://doi.org/10.1017/S1755048317000347 q92b (the death penalty for people who leave the Muslim religion) 0.6646 q92c (punishments like whippings and cutting off of hands for crimes like 0.8137 theft and robbery) q92d (stoning people who commit adultery) 0.8305 , on Religiosity q59 (How much, if at all, does the way you live your life reflect the Hadith 0.4611 28 Jun2017 at15:38:25 and Sunna) q61 (prayer frequency) 0.7926 q65 (listening/reading Quran frequency) 0.7664 q64e (Do you give Zakat) 0.7199 q64f (fast during Ramazan) 0.6758 Secularism q15 (How much influence should religious leaders have?) 0.4188

, subjectto theCambridgeCore q68/69 (How much do the laws follow Sharia?/ Is this a good or a bad 0.8187 thing)

Note: Small loadings (i.e., less than 0.4) are left as blank. 20 Fair, Hamza, and Hell

from zero to one, is an additive index derived from five questions (detailed in Appendix 1) that tap into respondent support for Quranic literalism and physical punishments. Factor analysis (Table 2) confirms the two distinct concepts comprising this index. To test H5, we operationalize the notion that Sharia provides service using a question that asks respondents whether or not they believe Muslims leaders should decide family and property disputes. Fair, Littman, and Nugent (2017) had access to a survey that they specifically designed to test their hypotheses about the different ways respondents conceptualize Sharia and their beliefs about Islamist extremism. Unfortunately, we must make do with the questions afforded by Pew how- soever, imperfect. We believe that this question about the role of religious leaders in adjudicating family and personal disputes is the best way to operationalize this posited relationship put forth in their study because recourse to religious leaders as an informal and timely means to resolve these conflicts help preclude extrajudicial violence in Bangladesh (and elsewhere in South Asia), particularly given the enormous backlog in Bangladesh’s judicial system (United Nations Development Program 2013). To test H6, we created a third additive index variable to proxy support for respondent secularism based upon three questions (described in Appendix 1). The first asked how much influence religious leaders should have in political affairs. The second and third questions asked respondents whether they believed Bangladesh follows Sharia law and whether or not they believe this is a good or bad thing. This index ranged from zero (least supportive of secularism) to one (most supportive). Factor analysis confirmed the intuition behind this index as well, as shown in Table 2. Based upon previous work by Shafiq and Sinno (2010) among others who have studied respondent-level support for Islamist violence, we included the following control variables: the respondent’s level of education, gender, and age. We provide descriptive statistics for the dependent, independent, and control variables in Tables 1a and 1b. Note that we indicate the reference categories that we employ in our regression models with an “*” in Table 1a. We estimated three models for our dependent variable (support for suicide bombing), using ordinary least squares regression (OLS).10 Unless noted otherwise, we treat “Don’t Know/Refused” responses as missing. Table 1a provides the response categories (including “Don’t Know/Refused) for each of the variables we employ. We perform all anal- yses using the sample weights provided by Pew, as recommended in Pew’s

documentation. These weights are important because, per Pew’s methodology, Pew’s sample has an urban bias. Pew uses these weights to adjust for this fundamental bias in sample collection. Our OLS results are available in Table 3. The first model includes only the Islamism, piety, and secularism independent variables to isolate their effects whereas in the second model, we included only the economic variables to isolate their impacts. The third model included the entire set of independent and control variables. We consider Model 3 to be the complete model.

5.DISCUSSION OF EMPIRICAL RESULTS

As the summary data in Table 1a indicate, 47% of our sample justifies suicide attacks in some measure. In Table 4, we benchmark this level of observed support relative to other national Muslim populations surveyed

Table 3. Regression results (with weights)

(1) (2) (3) suicide_bombing suicide_bombing suicide_bombing b/se b/se b/se

Hudood 0.430*** 0.398*** (0.098) (0.099) Provides −0.130** −0.130** (0.053) (0.054) Piety 0.139 0.195 (0.151) (0.155) Secularism −0.076 −0.087 (0.115) (0.115) Econ_perceived 0.136*** 0.124*** (0.031) (0.031) Econ_actual −0.001 0.007 (0.010) (0.011) Educ −0.029*** (0.011) Male 0.115** (0.049) Age −0.000 (0.002) Constant 1.598*** 1.480*** 1.231*** (0.134) (0.114) (0.183) N 1863 1832 1826 R2 0.0136 0.0108 0.0293 Dep Var Mean 1.84 1.85 1.85

*p < 0.10; **p < 0.05; ***p < 0.01. Source: In-house analysis.

Table 4. Levels of support for suicide bombing over time

Muslims in… 2002 2004 2005 2006 2007 2008 2009 2010 2011 2013 2014 %%%%%%%%%%% . IPaddress: Bangladesh 44 ——————47 ——47 Palestinian Territories ————70 — 68 — 68 62 46 Lebanon 74 — 39 — 34 32 38 39 35 33 29 73.180.230.14

. ——— https://doi.org/10.1017/S1755048317000347 Egypt 288 131520282524 Turkey 13 15 14 17 16 3 4 6 7 16 18 Jordan 43 — 57 29 23 25 12 20 13 12 15 Tunisia —————————12 5 , on Malaysia ————26 ————27 18 28 Jun2017 at15:38:25 Indonesia 27 — 15 10 10 11 13 15 10 6 9 Pakistan 33 41 25 14 9 5 5 8 5 3 5 Tanzania 18 ———11 12 ————26 Nigeria 47 ——————34 — 819 Senegal 28 ————————18 15 Israel ——————7 — 20 7 16

, subjectto theCambridgeCore Note: Asked of Muslims only. Source: Compiled from Pew Research Center (2014; 2013a; 2013b; 2002). Who Supports Suicide Terrorism in Bangladesh? 23

by Pew from 2002 to 2014. Pew asked this question in Bangladesh only in 2014, 2010, and 2002. In 2014 and 2010, support for suicide attacks was highest in Bangladesh. Historically, from 2002 to 2014, the highest level of support for suicide attacks came from the Palestinian Territories (70% in 2007) and Lebanon (74% in 2002). As noted above, this question about support for suicide attacks is problematic. Support for the tactic, as discussed, has been shown to vary when respondents are given different versions of this question that vary details about the attacker, the victim and the political context in which the tactic is used. For this reason, it is important to note that Bangladesh has experienced only four suicide bombings as of 2016 (Chicago Project on Security and Threats 2016) compared to other countries surveyed by Pew, which have experienced such attacks more frequently. Moreover, the suicide attacks in Bangladesh have had some of lowest victim yields observed in suicide bombings (Chicago Project on Security and Threats 2016). (See data table in Appendix 2 for details.) It is possible that support for suicide attacks in Bangladesh is high because it has witnessed so few attacks. In Pakistan, for example, support for suicide bombing began to decline once Pakistanis began expe- riencing the lethality of the tactic in their own country (Fair 2009). This suggests a free-rider effect: persons may be willing to support this kind of violence as a show of force against the “enemies of Islam” (as the question is phrased), provided that their own country does not experience the negative consequences of its use. Turning to the results of our model estimations, we find no evidence from either model 2 or 3 to support H1 that economic status varies with support for Islamist violence, as some other afore-cited studies have found. Because the variable for this measure is problematic for several reasons previously explained, we cannot rule out measurement error with this variable. To test H1 more thoroughly, we require a better measure of income than we have in this survey. However, we find strong evidence for H2, which holds that perceived economic status will positively vary with support for Islamist violence. This is consistent with the findings of Blair et al. (2013) who found that wealthier Pakistanis tend to be more supportive of Islamist violence. They found that the urban poor were in fact the most opposed to Islamist militant groups. By using district-level violence as an explanatory factor for public support for Islamist militants, they conclude that one reason why the urban poor are most opposed to these groups is that they are far more likely to experience the negative externalities of this violence relative

to wealthier Pakistanis as Pakistani terror groups tend to conduct attacks in congested areas in Pakistanis cities, where the wealthy are less likely to live and conduct business. We are unable to replicate this approach here because Pew does not provide locational information. This is likely an important future direction for research should suitable data become available. Consistent with our prediction in H3, we find no statistically significant relationship between piety and support for suicide bombing in either model 1 or 3. This finding for Bangladesh is similar to the finding of Fair, Littman, and Nugent (2017) for Pakistan. Turning to H4 which posited that those who support scriptural literalism will be more likely to support Islamist political violence, we find strong support for this hypothesis in models 1 and 3. This result also comports with that of Fair, Littman, and Nugent (2017). Unravelling why such persons are more supportive of suicide attacks is a challenge given the suboptimal phrasing of Pew’s suicide bombing question and the fact that suicide bombing in Bangladesh is recent and rare. One possible explanation for this finding is that the Islamist militant organizations which have perpetrated suicide bombing, ostensibly to protect Islam, generally espouse such literalist interpretations of Islam and wage their campaigns of violence with the explicit goal of establishing a regime which is gov- erned by their version of Sharia. These groups include transnational organizations such as Al-Qaeda and the IS (and their regional affiliates) as well as regional Islamist militant groups such as, inter alia, the Afghan Taliban in Afghanistan, the Pakistani Taliban in Pakistan, al Shabab in Somalia, Hamas and numerous other groups in the Palestinian Territories. Second, while most of Bangladesh’ indigenous terrorist groups have not engaged in suicide bombing, they too have conducted violence with the explicit goal of imposing literalist interpretations of Sharia upon Bangladeshis (Roul 2016a). One recent study by Ciftci, O’Donnell, and Tanner (2017) supports this explanation in some measure. They utilized 2012 survey data for Muslim respondents from the Pew Global Attitudes Project surveys for Egypt, Jordan, Tunisia, and Turkey. They found that respondents “who desire the implementation of Islamic law based on a highly distinct interpretation of religious texts (literalist outlook) lean more favorably toward al-Qaeda” (Ciftci, O’Donnell, and Tanner 2017, 2). Turning to H5, we found no significant correlation between support for the traditional role of Islamist leaders resolving disputes and support for suicide bombing (H5). Recall that Fair, Littman, and Nugent (2017) did

find a correlation between these variables in their survey of Pakistan. There are several potential explanations for this discordant finding. First, the differences may simply be attributed to the different way in which both studies instrumentalized the dependent variables and/or the independent variables or other important differences in how the two studies were conducted, including the surveys that are employed in each. A second explanation may be the important differences in Bangladeshi and Pakistani polities and/or the kinds of Islamist actors operating in both countries. Unfortunately, we are unable to definitively account for these differences with these data. Among the control variables, two are significant: males are more likely to support suicide bombing as are persons with less education. Few studies have explored the impacts of gender upon support for Islamist violence. Those studies that have examined gender as an explanatory factor have found that gender’s effects vary widely across Muslim polities (Fair and Shepherd 2006). Similarly, with respect to education, Shafiq and Sinno (2010) find that the effect of educational attainment on support for suicide attacks varies across countries.11

CONCLUSION AND IMPLICATIONS

Bangladesh is a highly-populated and troubled country but it has generally been ignored by the security studies community and is under-represented in large-n surveys of the type employed in this study. This remains the case despite the variegated history of Islamist violence and the recent spate of violent attacks on non-Muslim minorities, Sufis, Ahmedis, liberal and secular social activists, and foreigners as well as the high levels of support for suicide bombing among the population. Even though both the IS and Al-Qaeda in South Asia have locked their sights on one of the world’s largest Muslim countries, the government remains rooted in a posture of denial about the activities of these two organizations on Bangladeshi soil. Equally disturbing, the country’s institutions are highly politicized and the two mainstream parties seem more interested in fighting each other than the serious sources of instability that threatens the country. There is a pressing need for more and better data about this important country. As we have discussed herein, these Pew data are important because they exist and are available to scholars. However, Pew uses sample sizes that are smaller than desired, poses questions that are not

specific to Bangladesh, and does no collect geographical reference points that would permit scholars to merge in other forms of data that may help explain variation in support for suicide attacks. Critically, due to the nature of these data, we are unable to study important potential relationships between geographical variation in exposure to violence and beliefs about violence. Despite the limitations of this present study, we believe that our findings are important if disquieting. Variables that strongly predict support for suicide bombings, such as our “Hudood” index, have significant support. We suspect — but cannot show definitively — that this may be due to the fact that major transnational and regional Islamist militant groups engaging in suicide bombing seek to establish emirates wherein they impose their interpretations of Sharia as do the various Bangladeshi domestic militant groups. And as we noted earlier in this paper, Bangladesh has witnessed a long and often sanguinary political struggle between those who favor secularism and those, such as the BJeI, who seek to transform Bangladesh into an IS While popular support for notions of Islamist leaders providing dispute resolution is high, with some 70% of respondents embracing it, the miti- gating value of this variable is less than the exacerbating contribution of the “Hudood” index. Bangladesh has consistently attracted the attention of development economists and sociologists and it has long been the recip- ient of developmental aid intended to expand employment and educational opportunities. However, these variables are opposing in their impacts upon support for suicide terrorism. While education seems to mitigate support for violence, higher perceived economic standing exacerbates it. The different influences of these variables suggest that developmental assistance is unlikely to have the palliative effects aid organizations often claim. Too often scholars, analysts, and policy makers find themselves chasing problems only after they emerge. There is a real chance to prevent both the further spread of Islamist militancy and increased lethality in the attacks in this important but overlooked country. This can happen only if scholars and policy makers embrace this challenge earnestly and rigorously. We hope that this present study motivates others to begin making Bangladesh a focus of crucial scholarly and policy analytic inquiry.

NOTES

1. About 11% of Bangladesh’ total population of 169 million is non-Muslim, the vast majority of which are Hindu.

2. As in April 2016, it was the fourth largest contributor with some 7,298 personnel involved in peacekeeping operations. United Nations. 2016. “Ranking of Military and Police Contributions to UN Operations.” www.un.org/en/peacekeeping/contributors/2016/apr16_2.pdf (Accessed on May 13, 2017). 3. Between 1988 and 2015, Bangladesh’s military expenditures total some $38 billion (in constant 2014 dollars). In terms of military expenditures between 1988 and 2015, it ranks 105 out of 167 mil- itaries for which the Stockholm International Peace Research Institute has data (N.d.). 4. The stateless Rohingyas comprise an ethnic group that is overwhelmingly Muslim and reside on lands that straddle the border between Myanmar and Bangladesh, neither of which admits them as citizens. 5. Ostensibly due to its deep involvement with terrorism and, given the state of judicial independence, likely due to BJeI’s alliance with the BNP, Bangladesh’s Supreme Court cancelled its registra- tion from the Election Commission, prohibiting it from contesting the 2014 general elections (New York Times 2013). BJeI has appealed this decision and awaits adjudication by a full bench. In the meantime, the party remains a legitimate actor in the country. 6. Fair (2015) using data from Pakistan, found that sectarian commitment was a strong predictor of whether or not a person would support specific militant groups. She argued that this correlation likely stemmed from the fact that Pakistan’s various militant organizations draw upon particular sectarian traditions within Sunni and even Shia Islam. 7. Ciftci, O’Donnel, and Tanner (2015) use a similar when they make a similar distinction between what they call the “religiosity and secular/Islamist cleavage.” Using 2012 and 2013 data from Pew’s Global Attitudes Survey, they create a variable for “religiosity,” which includes respondents’ answers about prayer frequency, fasting and importance of religion. (This closely resembles our “piety” index.) They also create a variable that measures whether respondents think laws should be made according to the principles of Quran, individual concern for Islamic extremism and their views about the role of religion in politics. In this study, we separate out the support for Quranic literalism and support for secular politics. 8. To be certain, this is not an ideal measure of support for violence in part because it is very abstract. In 2006, WorldPublicOpinion.org and Search for Common Ground fielded simultaneous surveys of Iranians and Americans with the purpose of identifying divergences and convergences of opinions on key issues. Both populations were asked whether or not they believed “attacks inten- tionally aimed at civilians are justified.” Whereas 80% of Iranians said they were “never justified,” only 46% of Americans answered similarly. At first blush, this would suggest that Americans are more supportive of suicide attacks than are Iranians. However, when asked whether “attacks by Palestinians on Israeli Civilians” are justified, 41% of Iranians said they were “never justified” compared to 80% of American respondents. Clearly, the abstract question about generalized attacks on civilians does not characterize individual support for this kind of violence (Kull 2007). 9. It also draws from elements of the “secular/Islamist cleavage” variable used by Ciftci, O’Donnell, and Tanner (2015). 10. This is a common estimation strategy when the dependent variable has four or more response categories both because the dependent variable reasonably approximates a continuous variable and because interpretation of OLS results is considerably more straightforward than those from an Ordered Logit Regression. Note that we also estimated the three models using Ordered Logit Regression as well, which identified the same variables as being statistically significant and in the same direction. While we discuss the OLS estimates in the main paper, we have placed the Ordered Logit Regression in Appendix 3. 11. Shafiq and Sinno hypothesized that educational attainment should discourage support for violent attacks “because formal education should instill ordinary men and women with values and skills that reduce support for suicide bombing” (Shafiq and Sinno 2010, 149).

REFERENCES Abadie, Alberto. 2006. “Poverty, Political Freedom, and the Roots of Terrorism.” American Economic Review 96:50–56.

Abdul-Ahad, Ghaith. 2008. “Face to Face with the Taliban.” www.theguardian.com/world/ 2008/dec/14/afghanistan-terrorism (Accessed on May 13, 2017). Agrawal, Ravi. 2016. “Is Bangladesh the next ISIS hotspot?” www.cnn.com/2016/06/20/ asia/bangladesh-new-isis-hotspot (Accessed on May 13, 2017). Al Arabiya English. 2015. “ISIS Claims Attack on Ahmadi Mosque in Bangladesh.” www. english.alarabiya.net/en/News/world/2015/12/26/ISIS-claims-attack-on-Ahmadi- mosque-in-Bangladesh.html (Accessed on May 13, 2017). Al-Mahmood, Syed Zain. 2015. “Bangladesh Grapples with Islamist Militancy.” www. wsj.com/articles/bangladesh-islamists-gain-ground-1443981583 (Accessed on May 13, 2017). Barry, Ellen, and Julfikar Ali Manik. 2015. “Al Qaeda Claims Responsibility for Attacks on 2 Publishers in Bangladesh.” www.nytimes.com/2015/11/02/world/asia/bangladesh- publishers-attacked-al-qaeda-claims-responsibility.html?_r=0 (Accessed on May 13, 2017). Bashir, Farasha. 2015. “Rising Extremist Worries in Bangladesh.” www.thediplomat.com/ 2015/10/rising-extremist-worries-in-bangladesh (Accessed on May 13, 2017). Bass, Gary J. 2013. The Blood Telegram. New York, NY: Alfred A. Knopf. BBC News. 2016a. “Bangladesh Hindu Priest Beheaded ‘by Islamic State.’” www.bbc. com/news/world-asia-35626237 (Accessed on May 13, 2017). BBC News. 2016b. “Bangladesh Siege: Twenty Killed at Holey Artisan Bakery in Dhaka.” www.bbc.com/news/world-asia-36692613 (Accessed on May 13, 2017). Berrebi, Claude. 2007. “Evidence about the Link between Education, Poverty, and Terrorism among Palestinians.” Peace Economics, Peace Science, and Public Policy 13. Bhattacharyya, Bibhas. 2015. “Bangladesh Terrorist came to Bengal in 2005 to Set Up JMB Network.” (Accessed on May 13, 2017). Blair, Graeme, C. Christine Fair, Neil Malhotra, and Jacob N. Shapiro. 2013. “Poverty and Support for Militant Politics: Evidence from Pakistan.” American Journal of Political Science 57:30–48. Blair, Graeme, Jason Lyall, and Kosuke Imai. 2013. “Explaining Support for Combatants during Wartime: A Survey Experiment in Afghanistan.” American Political Science Review 107:679–705. Bueno de Mesquita, Ethan. 2005. “The Quality of Terror.” American Journal of Political Science 49:515–530. Bullock, Will, Kosuke Imai, and Jacob N. Shapiro. 2011. “Statistical Analysis of Endorsement Experiments: Measuring Support for Militant Groups in Pakistan.” Political Analysis 19:363–384. Burgoon, Brian. 2006. “On Welfare and Terror.” Journal of Conflict Resolution 50:176–203. Calvert, John. 2002. “The Islamist Syndrome of Cultural Confrontation.” Orbis 46:333–349. Chandran, Anurag. 2015. “Al Qaeda in the Indian Subcontinent: Almost Forgotten.” (Accessed on May 13, 2017). Chicago Project on Security and Threats. 2016. “Suicide Attack Database.” (Accessed on May 13, 2017). Chiozza, Giacomo. 2007. “Disaggregating Anti-Americanisms: An Analysis of Individual Attitudes toward the United States.” In Anti-Americanisms in World Politics, eds. Katzenstein, Peter J., and Robert O. Keohane. Ithaca, NY: Cornell University Press, 93–128. Chiozza, Giacomo. 2011. “Winning Hearts and Minds: The Political Sociology of Popular Support for Suicide Bombings.” www.exeter.ac.uk/media/universityofexeter/elecdem/

pdfs/giacomochiozzatraining/How_to_Win_Hearts_and_Minds.pdf (Accessed on May 13, 2017). Ciftci, Sabri, Becky O’Donnel, and Allison Tanner. 2015. “Anti-Americanism and Support for Violent Organizations in Muslim Societies.” Presented at the Midwest Political Science Association Meeting, Chicago. Ciftci, Sabri, Becky O’Donnel, and Allison Tanner. 2017. “Who Favors al-Qaeda? Anti- Americanism, Religious Outlooks, and Favorable Attitudes toward Terrorist Organizations.” Political Research Quarterly doi:10.1177/1065912917702498. Collins, Kathleen. 2007. “Ideas, Networks, and Islamist Movements: Evidence from Central Asia and the Caucasus.” World Politics 60:64–96. Dal Bó, Ernesto, and Pedro Dal Bó. 2011. “Workers, Warriors, and Criminals: Social Conflict in General Equilibrium.” Journal of the European Economic Association 9:646–677. Dodwell, Brian, Daniel Milton, and Don Rassler. 2016. The Caliphate’s Global Workforce: An Inside Look at the Islamic State’s Foreign Fighter Paper Trail. West Point, NY: Combating Terrorism Center. Dwibhashyam, Devitra. 2016. “Hindu Man Hacked to Death in Bangladesh, Isis Claims Responsibility.” (Accessed on May 13, 2017). Eaton, Richard Maxwell. 1993. The Rise of Islam and the Bengal Frontier, 1204–1760. Berkeley, CA: University of California Press. Esposito, John L. 2002. Unholy War: Terror in the Name of Islam. New York, NY: Oxford University Press. Esposito, John L., and John O. Voll. 1996. Islam and Democracy. New York, NY: Oxford University Press. Fair, C. Christine, and Bryan Shepherd. 2006. “Research Note: Who Supports Terrorism? Insights from Fourteen Muslim Countries.” Studies in Conflict and Terrorism 29:51–74. Fair, C. Christine, and Seth Oldmixon. 2015. “Think Again: Islamism and Militancy in Bangladesh.” www.nationalinterest.org/feature/think-again-islamism-militancy- bangladesh-13567 (Accessed on May 13, 2017). Fair, C. Christine, Neil Malhotra, and Jacob N. Shapiro. 2012. “Faith or Doctrine? Religion and Support for Political Violence in Pakistan.” Public Opinion Quarterly 76:1–33. Fair, C. Christine, Neil Malhotra, and Jacob N. Shapiro. 2014. “Democratic Values and Support for Militant Politics: Evidence from a National Survey of Pakistan.” Journal of Conflict Resolution 58:743–770. Fair, C. Christine, Rebecca Littman, and Elizabeth Nugent. 2017. “Conceptions of Shari`a and Support for Militancy and Democratic Values: Evidence from Pakistan.” Political Science Research Methods doi:10.1017/psrm.2016.55. Fair, C. Christine, Rebecca Littman, Neil Malhotra and Jacob N. Shapiro. 2016. “Relative Poverty, Perceived Violence, and Support for Militant Politics: Evidence from Pakistan.” Political Science Research and Methods. doi:10.1017/psrm.2016.6. (June 19). Fair, C. Christine, Rebecca Littman, Neil Malhotra, and Jacob N. Shapiro. 2016. “Relative Poverty, Perceived Violence, and Support for Militant Politics: Evidence from Pakistan.” Political Science Research and Methods www.journals.cambridge.org/ action/displayAbstract?fromPage=online&aid=10188983&fileId=S2049847016000066 (Accessed on May 13, 2017). Fair, C. Christine, Jacob S. Goldstein, and Ali Hamza. 2016. “Can Knowledge of Islam Explain Lack of Support for Terrorism? Evidence from Pakistan.” Studies in Conflict and Terrorism www.tandfonline.com/doi/abs/10.1080/1057610X.2016.1197692# abstract (Accessed on May 13, 2017).

Fair, C. Christine. 2009. “Pakistan’s Own War on Terror: What the Pakistani Public Thinks.” Journal of International Affairs 63:39–55. Fair, C. Christine. 2014. “Al Qaeda’s Re-launch in South Asia.” (Accessed on May 13, 2017). Fair, C. Christine. 2015. “Explaining Support for Sectarian Terrorism in Pakistan: Piety, Maslak and Sharia.” Religions 6:1137–1164. Ganguly, Sumit. 2016. “Bangladesh Accommodates Extremism, Spelling Danger for the Region.” (Accessed on May 13, 2017). Gates, Scott. 2002. “Recruitment and Allegiance: The Microfoundations of Rebellion.” Journal of Conflict Resolution 46:111–1130. Ginges, Jeremy, Ian Hansen, and Ara Norenzayan. 2009. “Religion and Support for Suicide Attacks.” Psychological Science 20:224–230. Global Fire Power. 2016. “Bangladesh Military Strength: Current Military Capabilities and Available Firepower for 2016 Detailed.” (Accessed on May 13, 2017). Gowen, Annie. 2014. “Bangladesh’s Political Unrest Threatens Economic Gains, Democracy.” (Accessed on May 13, 2017). Gurr, Ted R. 1970. Why Men Rebel. Princeton, NJ: Princeton University Press. Habib, Haroon. 2009. “Wanted Militant Arrested in Bangladesh.” www.thehindu.com/ todays-paper/tp-international/wanted-militant-arrested-in-bangladesh/article233434.ece (Accessed on May 13, 2017). Hasan, Mubashar. 2011. “Historical Developments of Political Islam with Reference to Bangladesh.” Journal of Asian and African Studies 47:161. Hasan, Mubashar. 2012. “The Geopolitics of Political Islam in Bangladesh.” Harvard Asia Quarterly 14:60–69. Horowitz, Donald L. 1985. Ethnic Groups in Conflict. Berkeley, CA: University of California Press. Huntington, Samuel. 1984. “Will More Countries Become Democratic?” Political Science Quarterly 99:193–218. Huntington, Samuel. 1993. “The Clash of Civilizations?” Foreign Affairs 72:22–50. Huntington, Samuel. 1996. The Clash of Civilizations and the Remaking of the World Order. New York, NY: Simon & Schuster. Hussain, Zakir. 2016. “8 Bangladeshi Men Detained under Singapore’s ISA for Planning Terror Attacks Back Home.” www.straitstimes.com/singapore/8-bangladeshi-men- detained-under-singapores-isa-for-planning-terror-attacks-back-home (Accessed on May 13, 2017). International Crisis Group. 2014. “Myanmar: The Politics of Rakhine State Asia Report.” www.crisisgroup.org/~/media/Files/asia/south-east-asia/burma-myanmar/261-myanmar- the-politics-of-rakhine-state.pdf (Accessed on May 13, 2017). Islamic State. 1437. “Among the Believers are Men: Abu Jandal al-Bangali.” Dabiq 14:1–68; www. clarionproject.org/docs/Dabiq-Issue-14.pdf (Accessed on May 13, 2017). Jalal, Ayesha. 2008. Partisans of Allah: Jihad in South Asia. Cambridge, MA: Harvard University Press. Khandake, Tamara. 2016. “The Islamic State’s Leader in Bangladesh Is Probably Canadian.” www.news.vice.com/article/the-islamic-states-leader-in-bangladesh-is-probably- canadian (Accessed on May 13, 2017). Khondker, Habibul Haque. 2006. “Civil and Uncivil Society in Bangladesh.” www. archive.thedailystar.net/2006/04/21/d60421150196.htm (Accessed on May 13, 2017). Krueger, Alan B., and Jitka Malekova. 2003. “Education, Poverty and Terrorism: Is there a Causal Connection?” Journal of Economic Perspectives 17:119–144.

Kull, Steven. 2007. “Public Opinion in Iran and America on Key International Issues.” www.pdfs.semanticscholar.org/5887/cf832170a726ce73205773c80b03bb7ed3a0.pdf (Accessed on May 13, 2017). Kumar, Nikhil, and Rishi Iyengar. 2016. “Here’s What You Need to Know about Growing Extremist Violence in Bangladesh.” www.time.com/4312152/heres-what-you-need-to- know-about-growing-extremist-violence-in-bangladesh (Accessed on May 13, 2017). Kumar, Sanjay. 2014. “Islamic Fundamentalists Terrorize Minorities in Bangladesh.” The Diplomat, January 15. Laqueur, Walter. 1999. The New Terrorism: Fanaticism and the Arms of Mass Destruction. New York, NY: Oxford University Press. Lintner, Bertil. 2002a. “Bangladesh: Breeding Ground for Muslim Terror.” www.atimes. com/atimes/South_Asia/DI21Df06.html (Accessed on May 13, 2017). Lintner, Bertil. 2002b. “Is Religious Extremism on the Rise in Bangladesh?” www. asiapacificms.com/articles/bangladesh_extremism (Accessed on June 16, 2016). Lintner, Bertil. 2003. “Bangladesh Extremist Islamist Consolidation.” www.asiapacificms. com/papers/pdf/faultlines_bangladesh.pdf (Accessed on May 13, 2017). Mahmood, Adil. 2016. “AQIS v IS: Battle for Bangladesh?” www.dhakatribune.com/ bangladesh/2016/apr/29/aqis-v-battle-bangladesh (Accessed on May 17, 2017). Manik, Julfikar Ali, and Ellen Barry. 2016. “Bangladesh Gay Rights Activist Is Hacked to Death.” www.nytimes.com/2016/04/26/world/asia/gay-rights-activist-in-bangladesh-is- hacked-to-death.html (Accessed on May 17, 2017). Manik, Julfikar Ali. 2016. “Bangladesh Arrests Three Men in Priest’s Beheading.” www. nytimes.com/2016/02/23/world/asia/hindu-priest-beheaded-in-bangladesh-isis-said-to- claim-responsibility.html (Accessed on May 17, 2017). Marshall, Andrew R.C. 2014. “Menace or Myth, Myanmar Frets over Rohingya Militant Group.” www.reuters.com/article/us-myanmar-rohingya-security-idUSKCN0JE0UZ 20141130 (Accessed on May 17, 2017). Mendelsohn, Barak. 2005. “Sovereignty under Attack: The International Society Meets the Al Qaeda Network.” Review of International Studies 31:45–68. Montero, David. 2010. “Pakistani Militants Expand Abroad, Starting in Bangladesh.” www.csmonitor.com/World/Asia-South-Central/2010/0805/Pakistani-militants-expand- abroad-starting-in-Bangladesh (Accessed on May 17, 2017). Mousseau, Michael. 2011. “Urban Poverty and Support for Islamist Terror: Survey Results of Muslims in Fourteen Countries.” Journal of Peace Research 48:35–47. Mustafa, Sabir. 2013. “Hefazat-e Islam: Islamist Coalition.” www.bbc.com/news/world- asia-22424708 (Accessed on May 17, 2017). Pew Research Center. 2002. “What the World Thinks in 2002.” www.pewglobal.org/2002/ 12/04/what-the-world-thinks-in-2002 (Accessed on June 4, 2017). Pew Research Center. 2013a. “The World’s Muslims Data Set: Background and Codebook.” www.pewforum.org/datasets/the-worlds-muslims (Accessed on May 17, 2017). Pew Research Center. 2013b. The World’s Muslims: Religion, Politics and Society. Washington, DC: Pew Research Center. Pew Research Center. 2014. “Levels of Support for Suicide Bombing over Time.” www. pewglobal.org/2014/07/01/concerns-about-islamic-extremism-on-the-rise-in-middle- east/pg-2014-07-01-islamic-extremism-11 (Accessed on May 17, 2017). Piazza, James A. 2007. “Draining the Swamp: Democracy Promotion, State Failure, and Terrorism in 19 Middle Eastern Countries.” Studies in Conflict & Terrorism 30:521– 539.

Rheault, Magali, and Dalia Mogahed. 2008. “Many Turks, Iranians, Egyptians Link Sharia and Justice.” www.gallup.com/poll/109072/many-turks-iranians-egyptians-link-sharia- justice.aspx (Accessed on May 17, 2017). Riaz, Ali, and C. Christine Fair. 2011. “Islamist Militancy in Bangladesh: Regional and Extra-regional Implications.” In Political Islam and Governance in Bangladesh. Abingdon: Routledge. Riaz, Ali. 2007. Islamist Militancy in Bangladesh: A Complex Web. Abingdon: Routledge. Riaz, Ali. 2016. Bangladesh: A Political History since Independence. London: I.B. Taurus. Robinson, Francis. 2004. Islam and Muslim History in South Asia. New Delhi: Oxford University Press. Roul, Animesh. 2016a. “How Bangladesh Became Fertile Ground for Al-Qa`Ida and the Islamic State.” www.ctc.usma.edu/posts/how-bangladesh-became-fertile-ground-for- al-qaida-and-the-islamic-state (Accessed on May 17, 2017). Roul, Animesh. 2016b. “Spreading Tentacles: The Islamic State in Bangladesh.” www. jamestown.org/single/?tx_ttnews%5Btt_news%5D=43510&tx_ttnews%5BbackPid% 5D=7&cHash=cd9ebf36776530f195e5afdbf211935b#.V2nP97grKUl (Accessed on May 17, 2017). Scott, James C. 1976. Moral Economy of the Peasant: Rebellion and Subsistence in South East Asia. New Haven, CT: Yale University Press. Shafiq, M. Najeeb, and Abdulkader H. Sinno. 2010. “Education, Income and Support for Suicide Bombings: Evidence from Six Muslim Countries.” Journal of Conflict Resolution 54:146–178. Singh, S. Binodkumar. 2016. “Bangladesh Extremist Spike-Analysis.” www. eurasiareview.com/ 03052016-bangladesh-extremist-spike-analysis (Accessed on June 16, 2016). Stark, Alexandra. 2015. “Bangladesh on the Brink: Between Terrorism and Democracy.” www.thediplomat.com/2015/09/bangladesh-on-the-brink-between-terrorism-and- democracy (Accessed on May 17, 2017). Stern, Jessica. 2003. Terror in the Name of God: Why Religious Militants Kill. New York, NY: HarperCollins. Stockholm International Peace Research Institute. N.d. “SIPRI Military Expenditure Database: 1988–2015.” www.sipri.org/databases/milex (Accessed on May 17, 2017). Swami, Praveen. 2016. “Dhaka Café Attackers were Lightly Armed, Shared Single Gun for Photos.” www.indianexpress.com/article/world/world-news/dhaka-terror-attack- bangladesh-dhaka-cafe-hostage-crisis-terrorism-dhaka-gunman-2896086/2016 (Accessed on May 17, 2017). Tessler, Mark, and Jodi Nachtwey. 1998. “Islam and Attitudes toward International Conflict: Evidence from Survey Research in the Arab World.” Journal of Conflict Resolution 42:619–636. Tessler, Mark, and Michael D. H. Robbins. 2007. “What Leads Some Ordinary Men and Women to Approve of Terrorist Attacks Against the United States?” Journal of Conflict Resolution 51:305–328. The Daily Star. 2009. “Laskar-e-Taiyeba Active in Bangladesh for 14 Years.” www. archive.thedailystar.net/newDesign/story.php?nid=97387 (Accessed on May 13, 2017). The Hindustan Times. 2009. “Nabbed Terrorist Talked Daily with Pak Boss on Phone: Bangladesh.” www.hindustantimes.com/world/nabbed-indian-militant-talked-daily- with-boss-in-pakistan-dhaka/story-ovlfiWumG22ghzY1UNDTSI.html (Accessed on May 17, 2017). The Indian Express. 2015. “A Look at Some Key Radical Islamist Groups in Bangladesh.” www.indianexpress.com/article/world/world-news/gunmen-attacked-a-shiite-mosque- in-bangladesh (Accessed on May 17, 2017).

The New York Times. 2013. “Bangladesh: Islamic Party Banned from Taking Part in 2014 Election.” www.nytimes.com/2013/08/02/world/asia/bangladesh-islamic-party-banned- from-taking-part-in-2014-election.html?_r=0 (Accessed on May 17, 2017). Titus, Murray T. 1959. Islam in India and Pakistan: A Religious History of Islam in India and Pakistan. New Delhi: Munshiram and Monoharlal Publishers. Transparency International. 2016. “Corruption Perception Index 2015.” www.transparency. org/cpi2015 (Accessed on May 17, 2017). U.S. Central Intelligence Agency. 2016a. “The World Factbook: Egypt.” www.cia.gov/ library/publications/the-world-factbook/geos/eg.html (Accessed on May 17, 2017). U.S. Central Intelligence Agency. 2016b. “The World Factbook: Iran.” www.cia.gov/ library/publications/the-world-factbook/geos/ir.html (Accessed on May 17, 2017). U.S. Central Intelligence Agency. 2016c. “The World Factbook: Saudi Arabia.” www.cia. gov/library/publications/the-world-factbook/geos/sa.html (Accessed on May 17, 2017). U.S. Department of State. 2010. “Secretary of State’s Terrorist Designation of Falah-i- Insaniat Foundation.” www.state.gov/r/pa/prs/ps/2010/11/151931.htm (Accessed on May 17, 2017). U.S. Department of State Office of the Coordinator for Counterterrorism. 2006. “Country Reports on Terrorism 2005.” www.state.gov/documents/organization/65462.pdf (Accessed on May 17, 2017). United Nations Development Program. 2013. “Summary Report on Court Services Situation Analysis.” www.undp.org/content/dam/bangladesh/docs/Projects/JUST/ Summary_Report_on%20Court%20Services%20Situation%20Analysis.pdf (Accessed on May 17, 2017). United Nations Development Program. 2015. “Human Development Report 2015.” www. report.hdr.undp.org (Accessed on May 17, 2017). United Nations. 2015. “Ranking of Military and Police Contributions to UN Operations.” www.un.org/en/peacekeeping/contributors/2016/apr16_2.pdf (Accessed May 13, 2017.). van Schendel, Willem. 2009. A History of Bangladesh. Cambridge: Cambridge University Press. Weinberg, Leonard, Ami Pedahzur, and Daphn Canetti-Nisim. 2003. “The Social and Religious Characteristics of Suicide Bombers and Their Victims.” Terrorism and Political Violence 15:139–153. Wiktorowicz, Quintan. 2005. Radical Islam Rising: Muslim Extremism in the West. Oxford: Rowman and Littlefield.

Table 1.1. Variables used to construct indices Variable Description Survey Questions Used Hudood This is a positive measure (i.e., Q79a: Do you favor or oppose Index missing values, don’t know and making the sharia, or Islamic law, refused would be treated as zero) the official law of the land in our to measure the support for country? (1 if in favor) Hudood laws. It is based on six survey questions and possible values range from 0 to 1 (higher value means higher support for Hudood laws) Q66: The sharia is the revealed word of God or the sharia is developed by men, based on the word of God (1 if sharia is the revealed word of God) Q67: Which is closer to you view: the sharia should be open to multiple interpretations or there is only one, true understanding of the sharia? (1 if only one understanding of sharia) Q92b: Do you favor or oppose death penalty for people who leave the Muslim religion? (1 if in favor) Q92c: Do you favor or oppose punishments like whippings and cutting off of hands for crimes like theft and robbery (1 if in favor) Q92d: Do you favor or oppose stoning people who commit adultery? (1 if in favor) Piety Index This is a positive measure (i.e. Q59: How much, if at all, does the missing values, don’t know and way you live your life reflect the refused would be treated as zero) Hadith and Sunna, that is, the to measure religiosity. It is based sayings and actions of the Prophet on five survey questions and — a lot, a little, not too much, or possible values range from 0 to 1 not at all? (1 if a lot) (higher value means a person is more religious) Q61: People practice their religion in different ways. Outside of attending religious services, do you pray several times a day, once a day, a few times a week, once a week, a few times a month, seldom, or never? (1 if several times a day)

Q64: Please tell me how often you read or listen to the Koran. Would you say every day, at least once a week, once or twice a month, a few times a year or never? (1 if every day) Q64e: Do you give Zakat (that is give a set percentage of your wealth to charity or the mosque)? (1 if yes) Q64f: Do you fast, that is avoid eating during the daytime, during the holy month of Ramadan? (1 if yes) Secularism This is a positive measure (i.e. Q15: In your opinion, how much index missing values, don’t know and influence should religious leaders refused would be treated as zero) have in political matters? A large to measure secularism. It is influence, some influence, not too based on 3 survey questions and much influence or no influence at possible values range from 0 to 1 all? (1 if no influence) (higher value means a person is more secular) Q68 and Q69: This part is based on two questions. In Q68, in respondents are asked if their country follows sharia law and Q69 asks if that’s a good thing. It’s 1 if respondents say that his country very/somewhat closely follows (Q68) and that is a bad thing (Q69) It is also 1 if respondents says that his country doesn’t follow sharia law (Q68) and it’s a good thing (Q69).

APPENDIX 1. CONSTRUCTING THE INDICES

Below we provide the survey questions that we used in the construction of the Hudood, Piety, and Secularism indices.

Q15: In your opinion, how much influence should [In Iran: religious figures; in all other countries: religious leaders] have in political matters? A large influence, some influence, not too much influence or no influence at all?

1 Large influence 2 Some influence 3 Not too much influence

4 No influence at all 8 Don’t know (DO NOT READ) 9 Refused (DO NOT READ)

Q59: How much, if at all, does the way you live your life reflect the Hadith and Sunna, that is, the sayings and actions of the Prophet — a lot, a little, not too much, or not at all?

1 A lot 2 A little 3 Not too much 4 Not at all 8 Don’t know (DO NOT READ) 9 Refused (DO NOT READ)

Q61: People practice their religion in different ways. Outside of attending religious services, do you pray several times a day, once a day, a few times a week, once a week, a few times a month, seldom, or never?

1 Several times a day 2 Once a day 3 A few times a week 4 Once a week 5 A few times a month 6 Seldom 7 Never 8 Don’t know (DO NOT READ) 9 Refused (DO NOT READ)

Q64e: Do you give zakat that is give a set percentage of your wealth to charity or the mosque?

1Yes 2No 8 Don’t know (DO NOT READ) 9 Refused (DO NOT READ)

Q64f: Do you fast, that is avoid eating during the daytime, during the holy month of Ramadan?

1Yes 2No 8 Don’t know (DO NOT READ) 9 Refused (DO NOT READ)

Q65: Please tell me how often you read or listen to the Koran. Would you say every day, at least once a week, once or twice a month, a few times a year or never?

1 Every day 2 At least once a week 3 Once or twice a month 5 A few times a year 6 Never 8 Don’t know (DO NOT READ) 9 Refused (DO NOT READ

Q66: I am going to read you two statements, please tell me which comes closer to your view, even if neither is exactly right.

1 The sharia is the revealed word of God. OR 2 The sharia is developed by men, based on the word of God. 3 Both (DO NOT READ) 8 Don’t know (DO NOT READ) 9 Refused (DO NOT READ)

Q67: Which is closer to you view: the sharia should be open to multiple interpretations or there is only one, true understanding of the sharia?

1 Should be open to multiple interpretations 2 One true understanding 3 Neither (DO NOT READ) 8 Don’t know (DO NOT READ) 9 Refused (DO NOT READ)

Q79a: Do you favor or oppose making the sharia, or Islamic law, the official law of the land in our country?

1Favor 2 Oppose 8 Don’t know (DO NOT READ) 9 Refused (DO NOT READ)

Q92: Do you favor or oppose the following? (READ LIST)

1Favor 2 Oppose 8 Don’t know (DO NOT READ) 9 Refused (DO NOT READ)

Q92b: The death penalty for people who leave the Muslim religion Q92c: Punishments like whippings and cutting off of hands for crimes like theft and robbery Q92d: Stoning people who commit adultery

NOTE: We created a new dummy variable using Q68 and Q69. This new variable was a positive measure of respondent’s secularity. It was 1 if respondent believed that

• Their country’s laws closely follow sharia (if Q68 is 1 or 2) AND it is a bad thing (if Q69 is 2) • Their county’s laws do not follow share (if Q68 is 3 or 4) AND it is a good thing (if Q69 is 1)

Otherwiseit was coded as zero. Q68: How closely, if at all, do the laws in our country follow the sharia? Would you say very closely, somewhat closely, not too closely or not at all closely?

1 Very closely 2 Somewhat closely 3 Not too closely 4 Not at all closely 8 Don’t know (DO NOT READ) 9 Refused (DO NOT READ) ASK IF ANSWER GIVEN (Q68 = 1, 2, 3, 4)

Q69: And is this a good thing or a bad thing?

1 Good thing 2 Bad thing 3 Neither (DO NOT READ) 8 Don’t know (DO NOT READ) 9 Refused (DO NOT READ)

To validate our indices, we conducted factor analysis (principle component analysis). Method: principal-component factors; Rotation: orthogonal varimax (Kaiser off); Number of observations = 1,918; Retained factors = 5; Number of params = 55.

Table 1.2. Principal component analysis

Factor Variance Difference Proportion Cumulative

Factor 1 1.86919 0.33678 0.1438 0.1438 Factor 2 1.53242 0.29773 0.1179 0.2617 Factor 3 1.23468 0.00499 0.0950 0.3566 Factor 4 1.22969 0.12829 0.0946 0.4512 Factor 5 1.10140 . 0.0847 0.5360

LR test: independent vs. saturated; chi2 (78) = 1914.68; Prob > chi2 = 0.0000. We have provided the rotated factor loadings (pattern matrix) in Table 2 in the body of the paper. Below is the Factor rotation matrix.”

Table 1.3. factor rotation matrix

Factor 1 Factor 2 Factor 3 Factor 4 Factor 5

Factor 1 0.777 0.4819 0.2826 0.2314 −0.1752 Factor 2 −0.6119 0.7009 0.3352 0.115 −0.0929 Factor 3 −0.1251 −0.1722 −0.2389 0.9288 −0.1869 Factor 4 −0.0572 −0.4881 0.8617 0.1116 −0.0588 Factor 5 0.0547 0.0923 0.0903 0.241 0.9604

APPENDIX 2. SUICIDE ATTACKS BY COUNTRY

Number of Average Persons Average Persons Country Suicide Attacks killed per Attack Wounded per Attack

Bangladesh 4 3.25 19.25 Egypt 30 9.57 37.53 Indonesia 11 23.09 77.55 Israel 114 6.32 44.72 Jordan 3 19.00 40.00 Lebanon 72 15.10 28.53 Nigeria 175 14.07 25.35 Palestinian Territory, 59 1.14 5.58 Occupied Tanzania 1 11.00 74.00 Tunisia 4 9.25 5.00 Turkey 43 9.05 49.88

Source: Chicago Project on Security and Threats (2016).

APPENDIX 3. ORDERED LOGIT REGRESSION RESULTS

(1) (2) (3) suicide_bombing suicide_bombing suicide_bombing b/se b/se b/se

hudood_index 0.601*** 0.559*** (0.181) (0.185) provides −0.235** −0.220** (0.098) (0.099) religiosity_index 0.164 0.286 (0.279) (0.289) secularism −0.092 −0.113 (0.213) (0.215) econ_perceived 0.258*** 0.236*** (0.058) (0.059) econ_actual −0.015 0.004 (0.019) (0.020) educ −0.068*** (0.021) male 0.227** (0.091) age −0.004 (0.004) cut1 Constant 0.372 0.668*** 0.880** (0.245) (0.212) (0.342) cut2 Constant 1.274*** 1.573*** 1.798*** (0.246) (0.214) (0.344) cut3 Constant 2.547*** 2.844*** 3.091*** (0.255) (0.224) (0.351) N 1863 1832 1826 R2 Dep Var Mean 1.84 1.85 1.85

*p < 0.10; **p < 0.05; ***p < 0.01.